[Uim] Hi there

[yusuke at cherubim.icw.co.jp]

Hi!

> I'm Dave Oftedal. I study Japanese and I'm an avid Linux user, and this 
> is the second IME mailing list for Linux that I sign on to. ;)
Thanks! We are happy to hear opinions like you who is a Linnux use and
studying Japanese.

> I just got uim-anthy running today, thanks to Tabata-san, and it's 
> working very well. Curious thing, though... It seems like anthy has no 
> backend server... That's a bit different from Canna and Wnn?!?
Yes. anthy is implemented as a library.

As an author of Japese kana-kanji conversion system, multiuser
client-server system incurs severe overhead on implementation.
 (1) Authentication and encryption of communiation *MUST* be required,
     it is hard to implement.
 (2) Even if the communication is totally secure, conversion system
     must have strict separation of each user's data in a process.
 (3) When (1),(2) is implemented, there may be some implementation of
     user's mistakes that leads crackers to get privilege of
     other user (often root).
#There may be mistakes in uim or anthy's code, but it
#does not mean priviledge escalation.
 (4) Complex input methods like Japanese kana-kanji conversion system
     stores per-user private date into disks. Such data should be accessed
     not only by the conversion system, but also other tools like
     dictionary manager must be able to access the data.
#Please see ~/.anthy/ , if you are using anthy.

Also from other point of view, use of client-server system only
for some complex input methods is not desirable.
 (1) Most of input methods in the world can be implemented by
     simple composition table.
 (2) Servers always use memory and other resources, even when
     you are not using input methods.
 (3) Input method developer must stop running server, when they
     want to test newer code.

Actually, On MS-Windows, MacOS and cellular phones, MacOS and most of
PDAs uses library based input methods. I think library and script
based input method is the sane way for most of languages including
some complex languages.

Regards,
--
 CHAOS AND CHANCE!
  Yusuke TABATA (yusuke at cherubim.icw.co.jp)